gkrishnan commited on
Commit
c0c3c1b
1 Parent(s): 22eaf04

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +0 -19
app.py CHANGED
@@ -7,25 +7,6 @@ from transformers import AutoTokenizer
7
  import pickle
8
  import os
9
  import shutil
10
- from langchain.document_loaders import BSHTMLLoader, DirectoryLoader
11
-
12
- git clone https://github.com/TheMITTech/shakespeare
13
-
14
- from glob import glob
15
- files = glob("./shakespeare/**/*.html")
16
-
17
- os.mkdir('./data')
18
- destination_folder = './data/'
19
-
20
- for html_file in files:
21
- shutil.move(html_file, destination_folder + html_file.split("/"[-1]))
22
-
23
- bshtml_dir_loader = DirectoryLoader('./data/', loader_cls = BSHTMLLoader)
24
-
25
- data = bshtml_dir_loader.load()
26
-
27
- with open("shakespeare.pkl", "wb") as fp:
28
- pickle.dump(data, fp)
29
 
30
  with open('shakespeare.pkl', 'rb') as fp:
31
  data = pickle.load(fp)
 
7
  import pickle
8
  import os
9
  import shutil
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10
 
11
  with open('shakespeare.pkl', 'rb') as fp:
12
  data = pickle.load(fp)